Quantitative genetics of traits related to disease resistance and effects of vaccination in Atlantic salmon (Salmo salar)

(1)

Tale Marie Karlsson DrangsholtPhilosophiae Doctor (PhD) Thesis 2011:39 Norwegian University of Life Sciences • Universitetet for miljø- og biovitenskap Department of Animal and Aquacultural SciencesPhilosophiae Doctor (PhD) Thesis 2011:39

Quantitative genetics of traits related to disease resistance and effects of

vaccination in Atlantic salmon (Salmo salar)

Kvantitativ genetikk for egenskaper relatert til sykdomsresistens og effekter av vaksinering i Atlantisk laks (Salmo salar)

Tale Marie Karlsson Drangsholt

ISBN 978-82-575-1002-2 ISSN 1503-1667

Norwegian University of Life Sciences NO–1432 Ås, Norway

Phone +47 64 96 50 00

www.umb.no, e-mail: postmottak@umb.no

Nofima - The Norwegian Institute of Food, Fisheries and Aquaculture Research, Osloveien 1,

NO-1430 Ås, Norway

www.nofima.no, e-mail: nofima@nofima.no

(2)

Quantitative genetics of traits related to disease resistance and effects of vaccination in

Atlantic salmon ( Salmo salar )

Kvantitativ genetikk for egenskaper relatert til sykdomsresistens og effekter av vaksinering i Atlantisk laks (Salmo salar)

Philosophiae Doctor (PhD) Thesis Tale Marie Karlsson Drangsholt Dept. of Animal and Aquacultural Sciences

Norwegian University of Life Sciences Ås 2011

Thesis number 2011: 39

(3)

PhD supervisors Bjarne Gjerde Nofima Marin

P.O. Box 5010, 1432 Ås, Norway

Department of Animal and Aquacultural Sciences Norwegian University of Life Sciences (UMB) P.O. Box 5003, 1432 Ås, Norway

Jørgen Ødegård Nofima Marin

Department of Animal and Aquacultural Sciences Norwegian University of Life Sciences (UMB) P.O. Box 5003, 1432 Ås, Norway

PhD Evaluation Committee

Associate professor Henk Bovenhuis

Wageningen University, Animal Breeding and Genomics Center P.O. Box 338, NL 6700 AH Wageningen, The Netherlands Dr. Nick Elliott

CSIRO Marine and Atmospheric Research

Castray Esplanade, GPO Box 1538, Hobart TAS 7000, Australia Professor Odd Vangen

Department of Animal and Aquacultural Sciences, Norwegian University of Life Sciences

(4)

Acknowledgements

This study was founded through The Research Council of Norway (grant 179009/S40), The Fishery and Aquaculture Industry Research Fund and PHARAMQ AS. Thanks to SalmoBreed AS for supplying all fish used in the experiments and additional data. Thanks to PHARMAQ for providing funding, vaccines and skilled personnel for evaluation of vaccine induced side effects.

I would like to thank my main supervisor Bjarne Gjerde for his skilful and thorough supervision. You enthusiasm has inspired me, and I am grateful for our interesting discussions and for your encouragement. I would also like to thank my co- supervisor Jørgen Ødegård for valuable inputs and fruitful discussions, and for helping me through the mysteries of the DMU-program. Thanks to Hans B. Bentsen for bringing in valuable perspectives and inputs and for supporting me. Thanks to co- authors and members of the project group Frode Finne-Fridell and Øystein Evensn for interesting discussions and suggestions. Thanks to staff at Nofima Sunndalsøra and Averøy where most of my experiments were carried out. A special thanks to Kjellrunn Gannestad for your skilful help with the experiments and for always answering my questions.

Thanks colleagues at Nofima and at the Department of Animal and Aquacultural sciences (IHA) at UMB. Thanks to my (former) officemate Carloz Lozano for all good talks and computer help. Thanks to PhD-students Bente, Siri, Katrine, Hilde H. and everyone else at the 3rd floor at IHA for all your support, interesting (and funny) discussions, and the much welcome coffee breaks. Thanks to group leader Kari Kolstad and everyone in the breeding and genetics group in Nofima for creating a great work environment filled with interesting discussions and humour. Thanks to Celeste Jacq for helping me improve my language. It has meant a lot to me to have great colleagues around me!

Thanks to my friends and family for your support and the good times we have had together away from work. Last but not least I would like to thank my dear husband Anders for all your love, patience, and support. You are not only my partner and best friend but also a great scientist – thank you for all your advice and interesting discussions. I am very grateful for having you by my side!

Ås, May 2011

Tale Marie Karlsson Drangsholt

(5)

(6)

SUMMARY

Disease resistance is of major importance to the fish farming industry as disease outbreaks have negative effects on the industry’s economy, its reputation and on fish welfare. Today, almost all fish in the Norwegian salmon industry are vaccinated against a number of diseases, while selection for increased resistance to specific diseases is based on survival of unvaccinated fish in challenge tests. The main aim of this doctorial thesis was to obtain a better understanding of how to select for increased resistance to furunculosis in Atlantic salmon and how this relates to side effects of vaccination, taking into account that most fish in the industry are currently vaccinated. Resistance to furunculosis (survival) was recorded by challenge testing fish from 150 families (unvaccinated and vaccinated fish). Vaccine-induced side effects (adhesions of internal organs and melanin deposits) were recorded on samples of the 150 families at three points in time: after three months in freshwater (high temperature), and six and 12 months after sea transfer.

The first objective was to estimate the magnitude of the genetic (co)variation in survival of unvaccinated and vaccinated Atlantic salmon challenged with A.

salmonicida, the bacteria causing furunculosis. The results showed a low genetic correlation between resistance to furunculosis in unvaccinated and vaccinated fish. The second objective was to estimate the magnitude of genetic variation of the negative side effects of vaccination. Intermediate heritabilities were obtained for adhesions and melanin deposits. However, the results also showed that an alternative vaccine reduced the side effects compared to the standard vaccine. The third objective was to estimate the magnitude of the genetic correlation between disease resistance, side effects of vaccination, and harvest body weight. These traits were not genetically correlated;

though a possible exception is harvest body weight and survival of vaccinated fish, where a weak and unfavorable correlation was reported.

Today’s breeding strategy of testing unvaccinated fish is optimal if the long term goal is a reduced need for vaccination. Selection based on vaccinated fish is likely to be the most effective short term strategy, as all fish in the industry today are vaccinated.

However, this strategy is not very relevant for furunculosis as the vaccine is highly effective. Vaccine-induced side effects (adhesions and melanin deposits) could be reduced through selective breeding, but it is likely to be more appropriate to focus on other measures such as vaccine development. Selection for increased disease resistance, vaccine-induced side effects, or harvest body weight are not expected to lead to unfavorable correlated responses in any of these traits, with the possible exception of survival of vaccinated fish and harvest body weight.

(9)

SAMMENDRAG

Forbedret sykdomsresistens hos Atlantisk laks er viktig for oppdrettsnæringen ettersom sykdomsutbrudd har negativ påvirking på næringens økonomi og omdømme, og på fiskens velferd. I dag vaksineres det aller meste av fisken mot en rekke sykdommer. Samtidig pågår det et avlsarbeid for økt sykdomsresistens basert på resultater fra smittetester med uvaksinert fisk. Hovedmålet med dette doktorgradsarbeidet var å få bedre forståelse for hvordan man bør selektere for økt resistens mot bakteriesykdommen furunkulose (forårsaket av bakterien A. salmonicida), og hvordan dette er relatert til bivirkninger av vaksinering. Resistens ble målt som overleving i smittetester hos fisk fra 150 familier. Vaksinebivirkninger, sammenvoksinger av organ i bukhulen og melaninflekker på organ og bukvegg, ble målt på et tilfeldig utvalg av fisk fra de 150 familiene på tre ulike tidspunkt: Etter tre måneder i ferskvann (høy temperatur) og seks og tolv måneder etter sjøutsett.

I smittetest med furunkulose ble det funne høyere genetisk variasjon for uvaksinert enn vaksinert fisk og en relativ lav genetisk korrelasjon mellom furunkuloseresistens i uvaksinert og vaksinert fisk. For sammenvoksinger og melaninflekker ble det funnet middels store arvegrader. En alternativ vaksine gav reduserte vaksinebivirkninger sammenlignet med standardvaksinen. Egenskapene sykdomsresistens, vaksinebivirkninger og slaktevekt ble funnet å ikke være genetisk korrelert, med et mulig unntak mellom slaktevekt og overlevelse av vaksinert fisk i smittetest hvor det ble funnet en svak, ugunstig korrelasjon.

Dagens avlsstrategi basert på smittetester med uvaksinert fisk er optimal hvis det langsiktige avlsmålet er å redusere bruken av vaksinering. Seleksjon basert på vaksinert fisk er likevel den optimale strategien på kort sikt ettersom all fisk i næringen vaksineres, men en liten aktuell strategi for furunkulose ettersom dagens vaksine mot furunkulose er svært effektiv. Vaksinebivirkninger (sammenvoksninger og melaninflekker) kan reduseres gjennom alvsarbeid, men det er mest sannsynlig mer hensiktsmessig å fokusere på andre tiltak som for eksempel vaksineutvikling. Seleksjon for økt sykdomsresistens og slaktevekt og reduserte vaksinebivirkninger forventes ikke å gi ugunstige korrelerte responser i noen av de andre egenskapene, men overlevelse av vaksinert fisk og slaktevekt kan være et unntak.

(10)

LIST OF PAPERS

Paper I:

Drangsholt, T.M.K., Gjerde, B., Ødegård, J., Fridell, F., Evensen, O., Bentsen, H.B.

Quantitative genetics of disease resistance in vaccinated and unvaccinated Atlantic salmon (Salmo salar L.). Heredity, available online.

Paper II:

Drangsholt, T.M.K., Gjerde, B., Ødegård, J., Fridell, F., Bentsen, H.B. Quantitative genetics of vaccine-induced side effects in farmed Atlantic salmon (Salmo salar).

Aquaculture, 318(3-4): 316-324

Paper III:

Drangsholt, T.M.K., Gjerde, B., Ødegård, J., Fridell, F., Evensen, O., Bentsen, H.B.

Genetic correlation among disease resistance, vaccine-induced side effects and harvest body weight. Manuscript

(11)

(12)

SYNOPSIS

(13)

(14)

1 Introduction 1.1 Background

Aquaculture is the fastest growing food-producing sector with a total production of 5.2 billion tonnes in 2008 (fish, shellfish and crustaceans) of which around 60% is produced in China. The annual aquaculture production of Atlantic salmon is about 1.5 million tonnes at a value of US$ 7.2 billion (FAO). Interestingly, nearly 100% of this production is based on genetically improved stock (Rye et al., 2010), to a large degree a consequence of the positive results of the first family breeding program for Atlantic salmon that started in Norway in the 1970’s (Gjedrem, 2010). Today Norway is the world’s largest producer of Atlantic salmon accounting for around 50% of the global production. Disease resistance is of major importance to the fish farming industry, as disease outbreaks have negative effects on both the industry’s economy and its reputation. In the first years of production in the 1970s, the breeding goal focused mainly on growth. In the 1990s disease resistance based on survival in challenge tests was included after substantial genetic variation was detected for survival to furunculosis (Gjedrem et al., 1991). Later, the viral diseases infectious salmon anaemia (ISA) and infectious pancreatic necrosis (IPN) were also included in the breeding goal (Gjedrem, 2010; Storset et al., 2007). Today, the breeding objective of most breeding programs of Atlantic salmon includes from one to three specific disease resistance traits in addition to growth and one or several carcass quality traits (Rye et al., 2010).

The use of antibiotics was very high in the late 1980s in the Norwegian salmon farming industry due to frequent outbreaks of diseases like furunculosis, vibriosis and cold water vibriosis, and there was a high demand for preventive measures. Selective breeding for increased disease resistance was therefore regarded as an important contribution for reducing mortalities due to specific diseases. At the same time, effective vaccines were developed and vaccination against furunculosis began in 1990 (Håstein et al., 2005).

Thus, both selection and vaccination can act as measures for preventing disease outbreaks; yet to what degree these strategies are complimentary or conflicting remains to be studied.

The following parts of the introduction will present relevant background information for this thesis. Firstly, the principal disease to be studied (furunculosis), the most common

(15)

testing methods for disease resistance (challenge tests), and general information about survival in the seawater phase will be presented. Vaccination is involved in most of the results of this thesis and an introduction to vaccination and associated side effects will also be given. Literature on genetic parameters related to disease resistance and other traits relevant for this study will then be presented, followed by an introduction to analysis methods and selection theory relevant for disease resistance, as special consideration is required when using challenge tests Lastly, the breeding goal and the structure of a typical breeding program and is presented, as these have a vital role for implementation of the results of this thesis.

1.2 The main disease and testing methods 1.2.1 Furunculosis

Furunculosis (classical or typical) is a bacterial disease caused by Aeromonas salmonicida. salmonicida and is present in most aquaculture areas in the Northern Hemisphere. In salmon farming, infection with A. salmonicida occurs most often in seawater, and mortalities of unvaccinated fish are generally large. Furunculosis is believed to have been introduced to Norway in 1985 from Scotland and has subsequently caused severe losses to the industry and led to heavy use of antibiotics (Gjedrem et al., 1991; Håstein et al., 2005). In the 1990s, an effective vaccine became available, and this proved to be successful in combating the disease. The vast majority of farmed salmon produced in Norway are vaccinated against this and other diseases, and furunculosis is therefore not considered a threat to the salmon industry today (Håstein et al., 2005). Substantial genetic variation in resistance to furunculosis in unvaccinated fish has been reported (e.g. Gjedrem et al., 1991) and an effective vaccine is available. Thus, furunculosis can serve as a good model for estimating the magnitude of the genetic variation in survival of unvaccinated and vaccinated fish, and of the genetic correlation between these two traits.

1.2.2 Challenge tests

Challenge tests of fish in a controlled environment are widely used to test for disease resistance. This allows for full exposure to the infection in question and generally results in high mortalities. Further, most challenge tests can be performed on small fish and thus the economical value of each individual is relatively low. Different methods of

(16)

infection can be used for challenge tests. A cohabitation infection model closely mimics natural outbreaks, as the cohabitant shedders are injected with the pathogen in question and subsequently spread the disease through the natural route of infection to the non- infected fish in the same tank (Nordmo, 1997). This method has been used for challenge tests with A. salmonicida for nearly 20 years (Gjedrem et al., 1991; Gjøen et al., 1997;

Ødegård et al., 2007b). Another challenge model is intra peritoneal (i.p) injection where the pathogen is directly administered to non-infected fish and will thus circumvent the anatomical barriers of infection (e.g., gills, mucus and skin). This model has been used for challenge tests with V. salmonicida (cold water vibriosis) which can not otherwise be tested for in fresh water (Gjedrem and Gjøen, 1995). Challenge tests with ISA were earlier performed using i.p injections (Gjøen et al., 1997), but lately a cohabitation model has been developed and taken into use for this disease (Wetten et al., 2007;

Ødegård et al., 2007b). Challenge tests can also be performed through immersion (bath challenges), where the pathogen is added to the water. This method also mimics to some extent natural outbreaks, and has been used for challenge tests with the IPN-virus (causing IPN) (Kjøglum et al., 2008; Storset et al., 2007). To assess genetic variation in disease resistance, the experimental infection should mimic the natural infection route as closely as possible. It is therefore believed that infection through cohabitants or immersion/bath challenges are most appropriate. Challenge tests are also widely used to test the effectiveness of vaccines.

1.2.3 Survival in the seawater phase

The annual production loss of farmed Atlantic salmon in the seawater phase is high in Norway (23% in 2008), while even higher losses are reported in Scotland (28%) and in particular Chile (43%) (Gullestad et al., 2011). Conversely, the Faroe Islands have managed to reduce their production losses to about 5-10% (Gullestad et al., 2011).

General survival of fish in aquaculture is important from economic, animal welfare and ethical strand points, and increased survival could both decrease the production cost and increase the income. It has been estimated that a one percent unit reduction of production losses could decrease the cost by 25 million NOK (3.2 million €) and increase gross income by 200 million NOK (25.5 million €) for the salmon farming industry in Norway (Gullestad et al., 2011). There are thus huge economic incentives for increasing the overall survival. However, selection for increased survival based on crude mortality does not seem to be effective, due in part to variable causes of death over time (Vehvilainen et al., 2008). Furthermore, field records of infectious diseases in

(17)

aquaculture are generally unavailable as the cause of death is rarely recorded. One of the reasons for this lack in records is due to poor methodology for determining specific causes of death. However, Aunsmo et al. (2008a) developed such a method for Atlantic salmon and showed that it is possible to obtain cause-specific mortalities from commercial grow-out farms; yet this requires frequent investigation of dead fish and thus a high demand for resources. The development of a national system for better health and disease recording has been slow (Gullestad et al., 2011). Further, fish in commercial production are not individually tagged, but individual information on cause of death could potentially be obtained from the breeding companies’ test fish as well as from the breeding candidates. Little information exists regarding the correlation between survival to specific diseases in challenge tests and survival in the field, especially for vaccinated fish. Such research is, to the best of my knowledge, currently limited to one study on furunculosis (Gjøen et al., 1997) and one study on IPN (Wetten et al., 2007) (see 1.4.2 Genetic correlations between traits).

1.3 Vaccination and side effects

In comparison to most vertebrates, the immune system of fish is poorly studied; but it is clear that fish also have an adaptive immune system. This is a prerequisite for effective vaccine development as the vaccine exposes the fish to a specific pathogen and consequently induces specific immunity that protects the fish against the pathogen in the future (Press and Jørgensen, 2002). The first commercially available vaccine for fish was developed in the 1970s for enteric redmouth disease and vibriosis, and today vaccines are available for a number of diseases and species (Sommerset et al., 2005). In farmed Atlantic salmon, vaccination is one of the most important tools to prevent outbreaks of a number of bacterial and viral diseases, and is largely responsible for a reduction in the use of antibiotics in the 1990s (Ellis et al., 1997; Gudding et al., 1999;

Håstein et al., 2005). The use of vaccination against vibriosis started in 1977 followed by cold water vibriosis in 1987 and furunculosis in 1990 (Håstein et al., 2005). Today, farmed Atlantic salmon are routinely vaccinated before sea transfer against a number of bacterial (furunculosis, vibriosis, cold water vibriosis, winter ulcer) and viral (IPN, ISA, pancreas disease (PD) diseases (McLoughlin and Graham, 2007).

(18)

Fish can be immunized by oral-, immersion- or injection vaccines, and injections are most common in commercial production (Gudding et al., 1999). For some diseases such as furunculosis, an oil-adjuvant is needed in the vaccine in order to obtain long-lasting protection, as other adjuvants (aluminium salt and glucans) have shown inferior success (Gudding et al., 1999; Midtlyng et al., 1996). Most i.p. vaccines protect against a number of diseases (polyvalent), and as oil-adjuvant is needed for furunculosis, other antigens are normally added to this vaccine e.g., a vaccine containing six different antigens (ALPHA JECT® 6-2 (PHARMAQ, Oslo, Norway)).

Although oil-adjuvant vaccines are effective; they can lead to vaccine-induced side effects. The side effects can be seen as adhesions between intra-peritoneal organs and melanin deposits on internal organs and on the abdominal wall (Midtlyng et al., 1996).

In the most severe cases these vaccine-induced side effects can lead to reduced fish welfare and downgrading of the fish slaughter quality (Midtlyng, 1996; Poppe and Breck, 1997). The vaccines can also reduce fish appetite and growth (Aunsmo et al., 2008b; Berg et al., 2006; Midtlyng and Lillehaug, 1998; Sørum and Damsgård, 2004).

Adhesions and melanin deposits are seen in almost all vaccinated fish, and in a study by Lund et al. (1997), adhesions proved to be a good marker for distinguishing between vaccinated and unvaccinated fish, whereas melanin deposits were a poor marker.

Adhesions and melanin deposits are associated with prolonged inflammation caused by persistent antigens from the vaccine (Mutoloki et al., 2004). The severity of vaccine- induced intra-abdominal adhesions may be affected by the time (of year) of vaccination, vaccine formulation, water temperature and the size of fish at vaccination (Aucouturier et al., 2001; Berg et al., 2007; Berg et al., 2006). It has also been shown that oil- adjuvant vaccines can provoke systemic autoimmune reactions; and that adhesions and granulomas seen in the abdominal cavity of vaccinated fish are related to both chronic inflammation, and autoimmune reactions (Haugarvoll et al., 2010; Koppang et al., 2008). Genetics appears to play a role in the development of vaccine-induced side effects. In a pilot study, intermediate heritabilities (h²= 0.18 - 0.19) were found for susceptibility to intra-abdominal adhesions (Gjerde et al., 2009).

(19)

1.4 Genetic parameters

1.4.1 Genetic variation in disease resistance

Genetic variation in disease resistance is a prerequisite for response to selection, and can be estimated from challenge test data. Substantial genetic variation has been found for several diseases in Atlantic salmon; with heritability estimates for survival in challenge tests ranging from 0.43 to 0.62 for A. salmonicida (Gjedrem et al., 1991; Gjøen et al., 1997; Kjøglum et al., 2008; Ødegård et al., 2007b); from 0.19 to 0.32 for ISA-virus (Gjøen et al., 1997; Ødegård et al., 2007b); and from 0.31 to 0.55 for IPN-virus (Kjøglum et al., 2008; Wetten et al., 2007). Recently, a QTL explaining a very large proportion of the genetic variation in IPN resistance has been detected in both Scottish and Norwegian populations of Atlantic salmon (Houston et al., 2009; Moen et al., 2009). Genetic variation has also been found in resistance to parasites such as sea lice (Lepeophtheirus salmonis) (Gjerde et al., 2011; Kolstad et al., 2005), flatworms (Gyrodactylus salaris) (Salte et al., 2010) and amoebae (Neoparamoeba perurans) (Taylor et al., 2009). The genetic relationship between resistance to different diseases, both bacterial and viral, seems to be neutral or slightly favorable (Kjøglum et al., 2008;

Ødegård et al., 2007b). Because of biosecurity, the survivors in the challenge tests cannot be used as breeding candidates, and therefore selection must be based on records of their full- and half-sibs (family selection). The documentation of response to selection for increased disease resistance in selective breeding programs is limited to the results by Storset et al. (2007) who reported a substantial response in survival to IPN based on challenge test data.

1.4.2 Genetic correlations between traits

Many examples of unfavorable genetic correlations among traits and unfavorable correlated selection responses can be found in selective breeding experiments and programs of livestock, particularly in traits not included in the breeding objective (Rauw et al., 1998). Hence, reliable estimates of genetic correlations between traits are important so that measures can be taken to prevent such unfavorable correlated responses when selecting for other traits. Growth until harvest size (measured as body weight at harvest size), and survival to specific diseases in challenge tests, are two important traits in the breeding goal of Atlantic salmon; and the magnitude of the genetic correlation between these two traits is of great importance. However, few reliable estimates of genetic correlations between growth and other traits are available,

(20)

with none showing strongly unfavorable correlations (Gjedrem et al., 1991; Standal and Gjerde, 1987). Vaccine-induced side effects are currently not commonly included in the breeding goal of Atlantic salmon, and Gjerde et al. (2009) found no evidence of unfavorable genetic correlations of vaccine-induced side effects with survival to furunculosis and ISA in challenge tests. However, unfavourable genetic correlations of harvest body weight with both adhesion scores (-0.45 ± 0.10) and melanin scores (-0.27

± 0.11) were reported. When considering the correlation between survival in challenge tests and survival after outbreaks in the field, a very large genetic correlation has been reported for furunculosis survival (0.95) (Gjøen et al., 1997). When it comes to the correlation between survival in challenge tests and survival after outbreaks in the field, a very high genetic correlation has been reported for survival of to furunculosis (0.95) (Gjøen et al., 1997). These results were in all likelihood based on fish vaccinated with a vaccine without an oil-based adjuvant, which gave a weak and transient protection against furunculosis (Arne Storset, pers. comm.), although this was not explicitly stated in the paper. A large genetic correlation between field and challenge test data was also found for survival to IPN in two year-classes, where the field data of one year class was from vaccinated fish (rg=0.78 ± 0.16) and the other from unvaccinated fish (0.83 ± 0.07) (Wetten et al., 2007). Both year-classes had a large degree of mortalities suggesting that the IPN-vaccine was not very effective.

1.5 Selection for improved disease resistance 1.5.1 Statistical models for analysing survival data

There are several ways of analyzing survival data from challenge tests, and the issue has recently been addressed in a review by Ødegård et al. (2011a). A linear mixed model is a simple model based on survival at termination of the test or at an overall 50% survival of the population; such models were used in some of the earliest studies (Gjedrem et al., 1991), seeming to predict breeding values accurately (Gitterle et al., 2006; Ødegård et al., 2006; Ødegård et al., 2007a). These models assume a normal distribution of data, even though the observations are binary (dead or alive). A mixed threshold model takes into account the binary distribution of data, and assumes that the binary trait is fully explained by an underlying continuous trait (liability). Heritabilities estimated with linear models and threshold models can therefore not be directly compared as they are calculated on two different scales, and heritabilities on the underlying scale (threshold

(21)

models) are generally greater than on the linear scale (linear models). However, ranking of families is often similar using linear and threshold models (Gitterle et al., 2006;

Ødegård et al., 2006; Ødegård et al., 2007a). An alternative to these cross-sectional models is longitudinal models, where the time of death is considered rather than the binary notation of “dead or alive” at a given point in time. Proportional hazard models, where risk of mortality is a function of time, have been suggested as an appropriate model for lifetime data (Ducrocq and Casella, 1996). The survival score model is an approximation of this model, where individuals are scored as dead or alive in different sub-periods (e.g. days) (Ødegård et al., 2011a); results have shown that these can give a slightly more accurate estimation of breeding values compared to simpler cross- sectional models (Gitterle et al., 2006; Ødegård et al., 2006; Ødegård et al., 2007a).

An important issue in challenge testing is when to terminate the test. Common practice is to terminate the test when the cumulative mortality reaches about 50%, as this maximizes phenotypic variance for a binary trait. It is assumed that all individuals are susceptible and will die given sufficient time. However, if a non-susceptible (or resistant) fraction exists amongst the survivors, this assumption is violated and may bias both the cross-sectional and longitudinal models if the test is terminated before mortalities cease. In a study by Salte et al. (2010), where Atlantic salmon were challenged with the parasite Gyrodactylus salaris, a small genetic correlation was estimated between time until death and survival at the termination of the test (when mortalities had ceased). Thus, severe re-ranking of families would have occurred if the test had been terminated at 50% mortality. Selection for disease resistance should ideally be for reduced susceptibility rather than for a greater time until death (endurance). If a non-susceptible fraction exists, it may be possible to select for this by continuing the challenge tests until mortalities naturally cease and when most susceptible fish are assumed to be dead. A cure model has been developed in order to account for both susceptible and non-susceptible fish in a challenge test, and such a model has been applied to survival data of Penaeus vannamei challenge tested with taura syndrome virus (TSV) (Ødegård et al., 2011b). In the aforementioned study, endurance and susceptibility were seemingly two different traits (genetic correlation did not significantly deviate from zero) (Ødegård et al., 2011b), thus selection for susceptibility rather than endurance will then be possible. Similar studies on other species and diseases are necessary, yet currently lacking.

(22)

1.5.2 Selection theory

The goal of selection is to obtain (favorable) genetic changes in the traits, which can be defined as:

interval generation

variation genetic

accuracy intensity

G= × ×

Δ

Fish are typically highly prolific species; therefore, large full- and half-sib family groups can be used for challenge tests. Survivors of the tests are considered potential disease carriers, and thus they are not themselves used as selection candidates.

However, records of their full- and half-sib can be used to select breeding candidates (i.e. family selection). As family selection relies solely on information from between- family variation and not within-family variation (half of the total genetic variance), this method reduces accuracy. The maximum accuracy that can be obtained (with large sib- groups) is thus √0.5, and the intensity of selection is reduced in family selection as selection is restricted to random individuals from the highest ranking families rather than the best individuals across families (Ødegård et al., 2011a). However, if the within- family variation could be exploited (i.e., by selecting surviving fish) both the accuracy and the intensity of selection would increase.

In order to facilitate individual selection, indirect selection based on immunological markers correlated to disease resistance has been proposed as an alternative to challenge test data. Fjalestad et al. (1993) showed that the genetic correlation between the maker traits and survival needs to be large in order to be more efficient than direct selection for survival. Attempts to find such markers have not been very successful, as low genetic correlations have generally been reported (Gjedrem and Thodesen, 2005; Lund et al., 1995).

1.6 Breeding goal and structure of breeding programs 1.6.1 Breeding goal

Selective breeding is often a relatively slow process due to long generation intervals (three-four years in Atlantic salmon), and the current breeding goal must therefore remain relevant for at least 10-20 years in the future. However, in order to keep up with changes in the industry and markets, the breeding goal has to be adjusted over time. It is therefore relevant to talk about long and short term breeding goals. Short term breeding

(23)

goals generally try to solve a present problem or demand, while the long term breeding goals try to foresee what can be important in the future. In general, the breeding goal of the nucleus has to be long term, while a narrower and short term breeding goal can be applied on the multiplier level.

The breeding goal consists of all traits directly or indirectly selected for in a selection program, and is primarily of an economic nature (e.g. maximum profit) for most farm animals. The breeding goal for Atlantic salmon in Norway typically includes traits related to growth, quality and health. Health generally includes less deformities and increased resistance toward specific diseases like furunculosis, ISA and IPN (Rye et al., 2010; www.aquagen.no; www.salmobreed.no). The breeding goal of Atlantic salmon was initially only focused on increased growth, and this led to a significant increase in this trait. Age at sexual maturation was the second trait to be included and disease resistance and quality traits followed (Gjedrem, 2010). Growth will probably be important both in the long and short term, while quality traits (e.g. filet fat) may change over time as consumer preference changes. Further, to some degree there is an optimal level for quality traits and other traits such as age at sexual maturation, thus suggesting that such traits may be less important in the breeding goal in the long term. Survival is generally considered a long term breeding goal, as this is crucial for production from economic, animal welfare and ethical points of view (Gjedrem, 2005).

The relative economic weight for each trait in the breeding goal must be considered, and a common approach in farm animals is to derive economic values using profit equations (Nielsen et al., 2011). However, in fish breeding a desired or restricted gain approach is most commonly used (Brascamp, 1984; Nielsen et al., 2011; Olesen et al., 2000). In this method, the breeding goal is to obtain a desired or restricted genetic gain for each included trait, and the weights are then derived in order to reach that goal.

1.6.2 Structure of breeding programs

Selective breeding programs for fish typically consist of several levels. Firstly, a breeding nucleus contains test fish and breeding candidates. Secondly, at the multiplier level, offspring of selected fish from the nucleus are reared to sexual maturation from which eyed eggs, fry, or fingerlings are supplied to producers for the grow-out stage (Gjerde, 2005). .For species with very high fecundity (e.g. Atlantic cod, Gadus

(24)

morhua), the multiplier level may be skipped; but for Atlantic salmon, a multiplier is needed in order to produce enough eggs and fry to supply the market (Skagemo et al., 2010). In order to meet the different demands of the producers, brood stock groups on the multiplier level can be especially selected for a single or a limited number of traits like disease resistance quality traits and growth; thus developing different products to suit the specific requirements of different producers. The broodstock groups are developed from a few families selected based on breeding values for the desired traits (e.g. disease resistance). The resulting females are then mated to either males from the breeding nucleus or males from the same or a different broodstock group, and the resulting eyed eggs or fry are sold to the grow-out producers. Many of the traits selected for cannot be recorded on the selection candidates (e.g. disease resistance), and family selection is then used both at the nucleus and the multiplier level. The multiplier level allows for higher selection intensity than can be practiced in the nucleus, as (long term) inbreeding is not a concern.

Recently, marker assisted selection has been implemented in some breeding programs of Atlantic salmon (Moen, 2010), following the discovery of a QTL explaining a very large fraction of the genetic (and phenotypic) variation in IPN-resistance in both Scottish (Houston et al., 2009) and Norwegian (Moen et al., 2009) populations. In particular, this offers opportunities to select for specific traits in the broodstock groups at the multiplier level. Currently, at least one breeding company offers fry from broodstock tested for the IPN-QTL (Moen et al., 2009; www.aquagen.no) and others plan to offer similar products (www.salmobreed.no).

(25)

2 Objectives

Today almost all fish in the Norwegian salmon aquaculture industry are vaccinated against a number of diseases, but the use of oil-adjuvant vaccines leads to vaccine- induced side effects. In parallel with vaccination, selection for increased disease resistance is performed for several diseases. However, selection is based on survival of unvaccinated fish in challenge tests. There is therefore a great need to establish the genetic variation of traits related to disease resistance and vaccination, and the correlation between them.

The objectives of the thesis were:

• To estimate the genetic correlation between survival of unvaccinated and vaccinated Atlantic salmon challenged with A. salmonicida, the bacteria causing furunculosis.

• To estimate the genetic variation in vaccine-induced side effects.

• To estimate the genetic correlations between survival (of both unvaccinated and vaccinated fish), vaccine-induced side effects and harvest body weight.

(26)

3 Material and methods

3.1 Family material

The fish used throughout this study were from 150 full sib families (the offspring of 85 sires and 150 dams) from the breeding nucleus of the Norwegian breeding company SalmoBreed AS. The families were produced in November 2006, and the first feeding of the families occurred during February 5 to April 17, 2007. Selection for increased resistance to furunculosis and ISA had been performed for one generation (on parents of fish used in this study), and the breeding goal also included growth, fillet fat and fillet colour.

3.2 Subsamples and traits

The main traits analyzed were survival, vaccine-induced side effects and body weight.

These traits were recorded on defined subsamples (Table 1) of 10-15 fish from each of the 150 families. Table 1 gives an overview of the groups, treatments, and in which papers they are included.

(27)

Table 1: Overview of traits, sample groups, treatments and in which papers they are included.

Trait/Group Challenge

agent Environment

and duration Type of

vaccine Time of

vaccination Paper

I Paper

II Paper

III Survival

Fur A. salmonicida Challenge test,

21 days N/A x x

ISA ISV-virus Challenge test,

22 days N/A x x

IPN IPN-virus Challenge test,

39 days N/A x

FUR-SV^† A. salmonicida Challenge test,

60 days Commercial October 2007 x x

FUR-RD A. salmonicida Challenge test,

60 days Reduced dose October 2007 x Vaccine-

induced side effects*

FW Freshwater,

3 months (17°C) Commercial October 2007 x

SW6 Seawater,

6 months Commercial March 2008 x

SW6-RV Seawater,

6 months Experimental March 2008 x

SW12

Seawater,

12 months Commercial March 2008 x x

Body weight

SW12 Seawater,

12 months Commercial March 2008 x

† FUR-SV group is named FUR-V in paper III

*Adhesions in the abdominal cavity and melanin deposits Vaccines are described in section 3.3 Vaccines.

3.3 Vaccines

All three vaccines used in this study (Table 1) contained mineral oil as adjuvant and were administered through i.p. injections. The vaccines included antigens of A.

salmonicida, Vibrio anguillarum serotype O1 and O2, Vibrio salmonicida, M. viscosa and IPN virus (protection against furunculosis, classical vibriosis, cold water vibriosis, winter ulcer and IPN, respectively). The commercial vaccine was ALPHA JECT® 6-2 (0.1 ml pr fish). The reduced dose vaccine contained 60% less A. salmonicida antigen than the commercial ALPHA JECT 6-2 and was administered with a reduced injection volume (0.05 ml pr fish). The experimental vaccine was administered with a reduced injection volume (0.05 ml pr fish) and is now a commercial vaccine (ALPHA JECT®

micro 6). All vaccines were produced by PHARMAQ. The vaccination time is given in Table 1.

(28)

3.4 Challenge tests

Challenge tests were performed at VESO Vikan (Namsos, Norway) between October 2007 and January 2008, and a cohabitation challenge model was used for all challenge tests. Challenge tests of unvaccinated fish (Fur, ISA, IPN) were part of SalmoBreeds routine genetic evaluation, and each of these diseases was tested in a separate tank. The vaccinated fish (Fur-SV and Fur-RD) were tested in another tank.

3.5 Vaccine-induced side effects

Vaccine-induced side effects were subjectively evaluated by trained personnel from PHARMAQ. Observations of the FW group were made at Nofima Marin Sunndalsøra in January 2008 and observations of the SW6, SW6-RD and SW12 groups occurred at Nofima marine, Averøy in November 2008 and June 2009. Adhesions were scored in three separate regions of the abdominal cavity (region 1, 2 and 3) using a scale from 0 – 6 and an interval of 0.5; where 0 = no adhesions and 6 = extremely severe adhesions (Midtlyng et al., 1996). Region 1 was anterior, region 2 was dorsal and posterior, and region 3 was the ventral part of the abdominal cavity. Melanin deposits on internal organs and the abdominal wall were scored on a scale from 0 – 3 with an interval of 1.0;

where 0 = no visible melanin and 3 = severe melanin spots.

3.6 Genetic parameter estimation

Estimates of heritabilities were obtained for survival (five traits), adhesions and melanin (each on four subsamples) and body weight (one trait). The genetic correlations between these traits were also estimated. Survival traits were analyzed using threshold models, and linear models were used for all other traits. The DMU software was used for all genetic analysis (Madsen and Jensen, 2007).

(29)

4 Main results

4.1 Paper I: Quantitative genetics of disease resistance in vaccinated and unvaccinated Atlantic salmon (Salmo salar L.) Substantial genetic variation was found in resistance to furunculosis in both the unvaccinated (heritabilities of 0.51 ± 0.05) and vaccinated (0.39 ± 0.06) fish. However, the genetic correlation between resistance to furunculosis in the two groups was low (0.32 ± 0.13), indicating a weak genetic association between resistance in the two groups. Significantly favourable genetic correlations were evident between resistance to furunculosis and resistance to both IPN and ISA in unvaccinated fish. This pattern was less pronounced for vaccinated fish.

4.2 Paper II: Quantitative genetics of vaccine-induced side effects in farmed Atlantic salmon (Salmo salar).

The SW6-RV group had lower adhesion scores (0.93 vs. 1.68) and melanin scores (1.04 vs. 1.49) than the parallel SW6-group. For the fish groups administered with the commercial vaccine, the heritability estimates for adhesion scores were 0.31 ± 0.05 (FW), 0.19 ± 0.04 (SW6) and 0.16 ± 0.05 (SW12), and for melanin scores 0.27 ± 0.05 (FW), 0.28 ± 0.05 (SW6) and 0.30 ± 0.05 (SW12). For the SW6-RV group, the heritabilities were lower; 0.08 ± 0.03 (SW6-RV) for the adhesion score and 0.11 ± 0.03 (SW6-RV) for the melanin score. The genetic correlation between adhesion scores and melanin scores within groups was intermediate for the FW group (0.52 ± 0.11) but higher for the SW6 (0.89 ± 0.06) and SW12 groups (0.87 ± 0.06). The genetic correlation between the SW6 and SW12 groups was large for both adhesion (0.89 ± 0.07) and melanin (0.92 ± 0.11) scores. However, the genetic correlations between FW group and the SW6 and SW12 groups were smaller for adhesions (0.62 ± 0.12 and 0.48

± 0.14, respectively) and for melanin (0.84 ± 0.08 and 0.61 ± 0.11, respectively). These results show that vaccine-induced side effects can be reduced through selective breeding, but that a reduction can also be achieved by other factors such as improvement of the vaccine.

(30)

4.3 Paper III: Genetic correlation among disease resistance, vaccine-induced side effects and harvest body weight

The genetic correlations between disease resistance traits and vaccine-induced side effects did not significantly deviate from zero. Likewise, the genetic correlation between body weight at harvest and vaccine-induced side effects did not significantly deviate from zero. Genetic correlations between harvest body weight and disease resistance (survival) were also weak; -0.18 ± 0.17 (furunculosis) and 0.05 ± 0.17 (ISA) for unvaccinated fish and -0.36 ± 0.16 (furunculosis) for vaccinated fish. These results suggest that selection for one trait is not likely to produce an unfavourable correlated response in the other traits, with a possible exception being for harvest body weight vs.

survival of vaccinated fish.

(31)

5 Discussion

The main aim of this thesis was to get a better understanding of how to select for increased resistance to furunculosis in Atlantic salmon and how this relates to side effects of vaccination, taking into account that most fish in the industry are currently vaccinated. Paper I focused on the magnitude of the genetic (co)variation in survival of unvaccinated and vaccinated fish in challenge tests, and found that survival from furunculosis in unvaccinated and vaccinated fish was weakly correlated genetically.

Paper II addressed the genetic variation of the negative side effects of vaccination and found that both adhesions and melanin deposits were heritable. Paper III examined the genetic correlation between disease resistance, side effects of vaccination and growth measured as body weight after 12 months at sea and found that these traits were generally not genetically correlated. The discussion will address issues relevant for selective breeding based on the above mentioned traits and results.

5.1 Challenge test data

Challenge tests are the most important tool when it comes to selection for improved disease resistance in aquaculture. However, as the surviving fish from these tests cannot be considered as breeding candidates due to biosecurity (fear of spreading the disease), the genetic variation within the family is not exploited. However, use of survivors as breeding candidates would also require measures to obtain records on these fish for other traits, especially other diseases. Therefore, use of genomic information is most likely a better approach to exploit the within family variation, either through marker assisted or genomic selection (Sonesson, 2007; Sonesson and Meuwissen, 2009).

However, implementation of marker assisted selection would require knowledge about more QTLs associated with resistance to disease (excepting IPN); and genotyping a large number of fish is still quite costly.

All challenge tests in this study (Paper I and Paper III) were performed by cohabitation.

This method mimics a natural outbreak, albeit at a greater infection pressure, and can therefore potentially capture genetic variation on both innate immunity (including anatomical barriers) and more specific immune reactions (adaptive immunity). One possible explanation for the positive genetic correlation found between the bacterial disease furunculosis and the viral disease ISA (Paper I) can thus be that some features

(32)

of innate immunity may be relevant for both diseases. Previous studies reporting neutral or negative genetic correlations have partially been based on methods (i.p.) circumventing the anatomical barriers (Gjøen et al., 1997; Kjøglum et al., 2008).

The vaccinated fish challenged in this study had mounted a specific immune response prior to the challenge, and a large pathogen load was needed in order to obtain sufficient mortalities. Innate immunity may thus play a less important role and the specific immunity a larger role in these fish, compared with the unvaccinated fish challenged in a different tank.

In our study, challenge tests with vaccinated fish (Fur-SV and Fur-RD) were run until a plateau was reached; whereas the tests with unvaccinated fish (Fur, ISA and IPN group) were terminated at an earlier point (close to 50% mortality), as this was common practice at the time of testing. The early termination of these tests may have affected the results to some extent, as it may obscure differences between potentially non- susceptible fish and susceptible fish still alive at the end of the testing period. However, it is more likely that a non-susceptible fraction would be most prominent in the vaccinated groups, and for these groups the testing period was continued until mortality naturally plateaued. Neither a cross-sectional nor a longitudinal survival model is expected to remove possible bias due to ignorance of potentially non-susceptible individuals. Furthermore, the two types of models did not give significantly different genetic correlations between survival of vaccinated and unvaccinated fish (Paper I).

Unvaccinated and vaccinated fish were tested in different tanks, and could alternatively have been tested in one common tank. However, in this scenario, the pathogen load would have to be very high to achieve sufficient mortality among the vaccinated fish, which may not be optimal for simultaneous testing of unvaccinated fish in the same tank.

5.2 Side effects of vaccination

Vaccination has been performed in parallel with selection for disease resistance, yet the vaccines used against furunculosis (oil-adjuvant vaccine) are known to cause side effects in terms of adhesions of internal organs and melanin deposits. Results from this study show that it is possible to select for less severe side effects, as significant genetic

(33)

variation was found for these traits (Paper II). The results also show that a vaccine with improved composition can effectively reduce the severity of the side effects (Paper II).

The latter result can be implemented immediately, whereas selective breeding is a long- term strategy. Hence, in my view it is unlikely that side effects of vaccination will be included in the breeding goal of commercial populations, as inclusion of yet another trait would necessarily lead to less gain in other more important traits. The genetic correlations between side effects and survival in challenge tests were found to be non- significant, and this was also found for side effects and body weight (Paper III). This shows that severity of these side effects is not expected to be effected by selection for increased disease resistance (in both vaccinated and unvaccinated fish) or selection for increased growth. Little genetic change in vaccine-induced side-effects should thus be expected as a result of selection on other traits.

5.3 Disease resistance in the breeding goal

Increased resistance (survival) to specific diseases is considered a long term breeding goal. However, vaccination in parallel with selection for increased disease resistance complicates this situation. The relatively weak (albeit favorable) genetic correlation between resistance to furunculosis in unvaccinated and vaccinated fish strongly indicates that there is a need to distinguish between a long and short term breeding goal for resistance to furunculosis and to determine what should be the main focus. A short term breeding goal would be to improve survival of vaccinated fish (as mass vaccination is likely to be required in the near future). However, through the current practice of vaccination, field mortality due to furunculosis is already very low (Håstein et al., 2005). The last three years showed no reported outbreaks of this disease in Norway (Veterinærinstituttet, 2008, 2009, 2010), but some unreported mortalities may exist. Improving genetic resistance to furunculosis in vaccinated fish would therefore have little additional practical value. However, if the aim is to make vaccination programs redundant in the future, a long term breeding goal should aim at improving furunculosis resistance of unvaccinated fish (i.e., as is the current practice for all diseases selected for). This selection strategy will lead to a gradual increase in disease resistance in the population, and vaccination of a decreasing fraction of the fish could be a potential strategy. However, partial vaccination has epidemiological implications and would need further investigation. Selection based on unvaccinated fish is expected to

(34)

give little, if any, observable improvement of field survival for the industry in the near future, due to the weak genetic correlation between resistance in vaccinated and unvaccinated fish and continued vaccination resulting in low prevalence of furunculosis in commercial salmon farming (due to vaccination).

Disease resistance is important in the breeding goal both from an economical and animal welfare point of view. However, how to assign economic weight to specific disease resistance traits as compared to other traits included in the breeding objective is a complicated issue. Desired or restricted gain is the most utilized method for determining economic weights for different traits in fish breeding schemes. If the long term breeding goal is a fish not in need of vaccination, disease resistance must be heavily emphasised in the breeding goal. The substantial heritability estimates for disease resistance reported in this study (Paper I) and several others (reviewed in Ødegård et al., 2011a) imply that selection for disease resistance may be effective.

However, as surviving fish cannot be considered as selection candidates, this will lower the genetic progress, and make the weight assigned to disease traits even more important. Even though health is becoming more important in the breeding goal of Atlantic salmon, growth is still heavily emphasised (30-50%) (Håvard Bakke, SalmoBreed, pers. comm.). Assigning relatively more weight to disease resistance will necessarily give less genetic gain for body weight (and other traits), even though there is no unfavourable genetic correlation between the traits (Paper III).

Disease outbreaks do not only cause big losses to the industry and effect profitability directly, but they are also of concern from an animal welfare perspective. Disease resistance thus has a non-market value in addition to a market value (Olesen et al., 2000), suggesting that the desired genetic gain for disease resistance should exceed the purely economic value of the trait. Selection on the multiplier level for a single or a limited number of traits (e.g. better disease resistance) is being practised by some breeding companies (e.g., AquaGen, SalmoBreed). This may be beneficial, as this can provide the market with fish with a higher genetic level of disease resistance without compromising too much on other traits (e.g., growth). A simulation study for fish species with a high fecundity showed that this could give substantial increased genetic gain for the grow-out producers as compared to random selection (Skagemo et al., 2010). This strategy enables breeding companies to supply customers with special

(35)

products, maintaining a stable and long term breeding goal in the nucleus. However, the additional genetic gain on the multiplier level is temporary (all offspring of selected broodfish will be slaughtered) while the gain in the nucleus is cumulative. Genetic gain obtained at the multiplier level would hence be limited to the gain in the breeding nucleus. The focus of the breeding nucleus therefore needs to be both on long term genetic progress and on maintaining genetic diversity.

Increased growth has been the main focus of selective breeding of Atlantic salmon, but disease resistance has been in included in the breeding goal for over 20 years (Gjedrem, 2010) and has become more important during the last few years. Unfavorable genetic correlations between growth and disease resistance would complicate selective breeding, but only low genetic correlations were found between growth and resistance to both furunculosis (unvaccinated and vaccinated) and ISA (Paper III). Only the genetic correlation between furunculosis resistance in vaccinated fish and growth was significantly different from zero, and selection for increased growth may thus lead to a slight decrease in disease resistance of vaccinated fish. However, this effect is unlikely to have any practical consequences, as the vaccine against furunculosis is very effective (Håstein et al., 2005; Veterinærinstituttet, 2008, 2009, 2010).

It has earlier been shown that vaccination can have a negative effect on harvest weight (Aunsmo et al., 2008b; Midtlyng and Lillehaug, 1998), but the reduced growth of vaccinated fish is not necessarily connected to the level of adhesions or melanin deposits; thus suggesting that reduced growth is a another side effect of oil-adjuvant vaccines (Aunsmo et al., 2008b). This is in agreement with results from Paper III, where small phenotypic and genetic correlations of growth with adhesions and melanin were reported; suggesting that genetic predisposition to growth does not affect the level of adhesions or melanin deposits. However, the new vaccine that reduced the level of adhesions and melanin also resulted in increased body weight (six months after sea transfer) (Paper II). A similar effect comparing two different vaccines with different injection volumes and from different manufactures has been reported earlier (Sørum and Damsgård, 2004). The negative effect of vaccination on body weight (unvaccinated and vaccinated with full and reduced dose) of 10-20% was found both at six and 12 months after sea transfer (Bjarne Gjerde, pers. comm.). This can be used as an argument to

(36)

assign more weight to disease resistance in the breeding goal, in order to reduce or avoid vaccination.

(37)

6 Concluding remarks The main findings from this study are:

Resistance to furunculosis in unvaccinated and vaccinated Atlantic salmon are only moderately positive genetically correlated and should therefore be looked upon as different genetic traits.

Selection for increased specific disease resistance based on vaccinated fish has no practical effect when an effective vaccine is available for the disease.

Therefore, if the long term breeding goal is to reduce or avoid vaccination, selection for increased resistance to furunculosis should be based on unvaccinated fish

Side effects of vaccination (melanin and adhesions) are heritable traits, which are not significantly genetically correlated to disease resistance or growth. Side effects of vaccination can be reduced through new vaccine formulation.

For unvaccinated fish resistance against furunculosis and ISA are both weakly genetically correlated with growth. Thus, selection for increased growth is not expected to result in unfavourable correlated responses in resistance to these diseases.

Resistance to furunculosis, ISA and IPN in unvaccinated fish were favourably genetically correlated, suggesting that some genetic variation exists for non- specific disease resistance in Atlantic salmon when all diseases are assessed using a cohabitation challenge model.

Based on these results my recommendation is continued selection for increased resistance to furunculosis based on unvaccinated fish, which in the long run may result in a population of fish where a decreasing fraction has to be vaccinated. However, in order to speed up this progress larger emphasis should placed on disease resistance in the breeding goal and thus less emphasis on other traits. Furunculosis is no longer considered a problem under farming conditions using vaccinated fish (Håstein et al., 2005), but efficient vaccines are still lacking for other diseases like IPN and PD (Biering et al., 2005). Similar studies with unvaccinated and vaccinated Atlantic salmon should be carried out for these diseases, and potentially also for other diseases and species where vaccination and selection are parallel prophylactic measures.

(38)

When it comes to vaccine-induced side effects (adhesions and melanin deposits) my recommendation is that these are not included in the breeding goal, but should be reduced through other measures such as vaccine development and optimization of vaccination procedures.

(39)

7 References

Aucouturier J, Dupuis L, Ganne V (2001). Adjuvants designed for veterinary and human vaccines. Vaccine 19(17-19): 2666-2672.

Aunsmo A, Bruheim T, Sandberg M, Skjerve E, Romstad S, Larssen RB (2008a).

Methods for investigating patterns of mortality and quantifying cause-specific mortality in sea-farmed Atlantic salmon Salmo salar. Dis Aquat Organ 81(2): 99-107.

Aunsmo A, Larssen RB, Valle PS, Sandberg M, Evensen Ø, Midtlyng PJ et al (2008b).

Improved field trial methodology for quantifying vaccination side-effects in farmed Atlantic salmon (Salmo salar L.). Aquaculture 284(1-4): 19-24.

Berg A, Rødseth OM, Hansen T (2007). Fish size at vaccination influence the

development of side-effects in Atlantic salmon (Salmo Salar L.). Aquaculture 265(1-4):

9-15.

Berg A, Rødseth OM, Tangerås A, Hansen T (2006). Time of vaccination influences development of adhesions, growth and spinal deformities in Atlantic salmon Salmo salar. Dis Aquat Organ 69(2-3): 239-248.

Biering E, Villoing S, Sommerset I, Christie KE (2005). Update on viral vaccines for fish. In: Midtlyng PJ (ed) Developments in biologicals Karger. Vol. 121, pp 97-113.

Brascamp EW (1984). Selection indexes with constraints. . Animal Breed Abstr 52:

645–654.

Ducrocq V, Casella G (1996). A Bayesian analysis of mixed survival models. Genet Sel Evol 28(6): 1-25.

Ellis AE, Gudding R, Lillehaug A, Midtlyng P, Brown F (1997). Immunization with bacterial antigens: Furunculosis. Dev Biol Stand 90: 107-116.

Fjalestad KT, Gjedrem T, Gjerde B (1993). Genetic improvement of disease resistance in fish: an overview. Aquaculture 111(1-4): 65-74.

Gitterle T, Ødegård J, Gjerde B, Rye M, Salte R (2006). Genetic parameters and

accuracy of selection for resistance to White Spot Syndrome Virus (WSSV) in Penaeus (Litopenaeus) vannamei using different statistical models. Aquaculture 251(2-4): 210- 218.

Gjedrem T (2005). Breeding plans. In: Gjedrem T (ed) Selection and Breeding Programs in Aquaculture. Springer: Dordrecht pp 251-277.

Gjedrem T (2010). The first family-based breeding program in aquaculture. Reviews in Aquaculture 2(1): 2-15.